Commits

Breadcrumbs

History for

onnasm-head

User selector

All users

All time

Commit History

Commits on Jan 14, 2017

• Add support for passes that run post-partitioning to OptimizationRegistry.

  Show description for 8eb161e
  

  hawkinsp

  authored and

  tensorflower-gardener

  committed
  8eb161e

  Copy full SHA for 8eb161e

Commits on Dec 1, 2016

• Don't remove the identity nodes that are introduced for control flow.

  Show description for a1c0bdf
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  a1c0bdf

  Copy full SHA for a1c0bdf

Commits on Nov 28, 2016

• Uses containers for per_step resources instead of a separate resource_mgr.

  Show description for b571f2e
  tensorflower-gardener

  committed
  b571f2e

  Copy full SHA for b571f2e

Commits on Nov 11, 2016

• Internal cleanup.

  Show description for 376307d
  tensorflower-gardener

  committed
  376307d

  Copy full SHA for 376307d

• Internal cleanup.

  Show description for 160f9b1
  tensorflower-gardener

  committed
  160f9b1

  Copy full SHA for 160f9b1

Commits on Nov 9, 2016

• Internal change.

  Show description for aac685b
  

  Dan Smilkov

  authored and

  tensorflower-gardener

  committed
  aac685b

  Copy full SHA for aac685b

Commits on Oct 25, 2016

• Internal cleanup.

  Show description for feebd9f
  tensorflower-gardener

  committed
  feebd9f

  Copy full SHA for feebd9f

Commits on Sep 20, 2016

• Adds support to print out the function body after various

  Show description for ebdd68f
  tensorflower-gardener

  committed
  ebdd68f

  Copy full SHA for ebdd68f

Commits on Aug 23, 2016

• Begin transition to use NodeDef in FunctionDef instead of

  Show description for e058f2d
  tensorflower-gardener

  committed
  e058f2d

  Copy full SHA for e058f2d

Commits on Aug 18, 2016

• Plumb the per-step resource manager through to the executor used when executing functions.

  Show description for ace4f5b
  tensorflower-gardener

  committed
  ace4f5b

  Copy full SHA for ace4f5b

Commits on Aug 11, 2016

• Move kernels defined in common_runtime/function.cc to a separate

  Show description for d6485c1
  tensorflower-gardener

  committed
  d6485c1

  Copy full SHA for d6485c1

• Using Session's environment in optimization functions

  Show description for bfb577a
  tensorflower-gardener

  committed
  bfb577a

  Copy full SHA for bfb577a

Commits on Aug 2, 2016

• Fix issue where the wrong version of the function library was used after partitioning.

  Show description for dff0e1b
  tensorflower-gardener

  committed
  dff0e1b

  Copy full SHA for dff0e1b

Commits on Jul 28, 2016

• Improve memory type deduction for operators with type list arguments.

  Show description for 34c820d
  tensorflower-gardener

  committed
  34c820d

  Copy full SHA for 34c820d

Commits on Jun 14, 2016

• In C++ shape inference, support registering a shape inference function in OP

  Show description for 72cd0b0
  tensorflower-gardener

  committed
  72cd0b0

  Copy full SHA for 72cd0b0

Commits on Jun 2, 2016

• Update copyright for 3p/tf/core.

  Show description for c8b59c0
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  c8b59c0

  Copy full SHA for c8b59c0

Commits on Jun 1, 2016

• Change function library to use a runner passed in from the Run call, instead of

  Show description for caed96e
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  caed96e

  Copy full SHA for caed96e

Commits on May 17, 2016

• Merge duplicate copies of reverse reachability pruning algorithm.

  Show description for fc7a945
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  fc7a945

  Copy full SHA for fc7a945

Commits on May 5, 2016

• Merge changes from github.

  Show description for 8bf6ef1
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  8bf6ef1

  Copy full SHA for 8bf6ef1

Commits on Apr 21, 2016

• Instead of just validating the memory type. Fix them if they are not

  Show description for 2c46a9e
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  2c46a9e

  Copy full SHA for 2c46a9e

Commits on Apr 2, 2016

• Enable constant folding in L0 optimization level.

  Show description for 0cd025b
  

  keveman

  authored and

  tensorflower-gardener

  committed
  0cd025b

  Copy full SHA for 0cd025b

Commits on Mar 24, 2016

• Allows the caller of a function to specify that a function is not

  Show description for 0cc25f0
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  0cc25f0

  Copy full SHA for 0cc25f0

Commits on Mar 23, 2016

• Allows explicitly setting a gradient function for a user defined function.

  Show description for f10637b
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  f10637b

  Copy full SHA for f10637b

Commits on Mar 18, 2016

• C++ gradients: Changes API to require specification of node outputs (also added some code cleanup).

  Show description for addfb7a
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  addfb7a

  Copy full SHA for addfb7a

Commits on Mar 16, 2016

• C++ gradients: split functions gradient code out into its own file.

  Show description for f4ade48
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  f4ade48

  Copy full SHA for f4ade48

Commits on Feb 26, 2016

• Add a step_id field to the OpKernelContext indicating which step

  Show description for e8260f3
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  e8260f3

  Copy full SHA for e8260f3

Commits on Feb 9, 2016

• Don't cache kernels in op segment if the op is stateless.

  Show description for 3d91f55
  

  A. Unique TensorFlower

  authored and

  keveman

  committed
  3d91f55

  Copy full SHA for 3d91f55

Commits on Feb 5, 2016

• Allows int32 on host memory work when a function passes int32 in/out.

  Show description for 0c6c4b9
  

  A. Unique TensorFlower

  authored and

  keveman

  committed
  0c6c4b9

  Copy full SHA for 0c6c4b9

Commits on Feb 3, 2016

• Refactor the logic to apply optimization into a common module.

  Show description for e830638
  

  A. Unique TensorFlower

  authored and

  Vijay Vasudevan

  committed
  e830638

  Copy full SHA for e830638

Commits on Feb 2, 2016

• Minor formatting fixes.

  Show description for a8d2f09
  

  Vijay Vasudevan

  authored and

  keveman

  committed
  a8d2f09

  Copy full SHA for a8d2f09

Commits on Jan 28, 2016

• Fixed constant folding to handle nodes with multiple outputs.

  Show description for c509c78
  

  keveman

  authored and

  Vijay Vasudevan

  committed
  c509c78

  Copy full SHA for c509c78

• Fixed constant folding to handle nodes with multiple outputs.

  Show description for bcd9722
  

  A. Unique TensorFlower

  authored and

  Vijay Vasudevan

  committed
  bcd9722

  Copy full SHA for bcd9722

• Fixed constant folding to handle nodes with multiple outputs.

  Show description for 33aa628
  

  keveman

  authored and

  Vijay Vasudevan

  committed
  33aa628

  Copy full SHA for 33aa628

Commits on Jan 26, 2016

• Change GraphDef versions to use version, min_consumer, min_producer

  Show description for a18d48f
  

  girving

  authored and

  Vijay Vasudevan

  committed
  a18d48f

  Copy full SHA for a18d48f

Commits on Jan 25, 2016

• Running our linter on a lot of files.

  Show description for 9b70316
  

  Vijay Vasudevan

  committed
  9b70316

  Copy full SHA for 9b70316